Translation Rates: Does Tool Use Affect the Price?

The most confusing thing about comparing translation quotations is that the same job can be priced at wildly different amounts. Part of that comes from differences in quality, and a good part of it from the way the calculation is made.

Contents

In this article we take up the link between translation rates and tool use through the verified findings of a case study carried out in Poland with 141 translators over the 2019 period. The figures belong to that sample and cannot be generalised to Türkiye.

Our aim is not to give a price list but to show which variables to look at when you are reading a quotation.

What unit are translation rates calculated on?

The first problem is not the price itself but the unit of measurement. Words, lines, characters and pages are all used side by side in the industry. When the same job is priced in different units, comparison becomes impossible.

The book gives a practical equivalence for this conversion: 1,800 characters with spaces correspond to roughly 240 words. This is the “standard page” measure common in the central European market, and the conversion has to be made before any comparison with word-based prices.

The confusion over units has a language side too. Whether volume is counted on the source language or on the target language produces a notable difference depending on the language pair. Counting on the target language means the final amount is not settled until the work is finished.

Tool use adds a further layer to this picture. In a CAT environment, volume is reported not as a raw word count but broken down into match categories. A file is therefore priced not with a single figure but with an analysis running to several lines.

The first effect of the tool on the rate, then, is that it changes the unit the rate is calculated on. Even where the price does not fall, the structure of the invoice changes completely.

There is an advantage to that change: the invoice becomes more transparent, and you can see which part is new translation and which part repetition. The disadvantage is that it makes comparison harder; two different discount scales produce different totals on the same file.

The difference in rate when translating into a foreign language

The clearest finding of the research on rates concerns the difference in direction. 80% of the participants who did not use a CAT tool said they applied a higher rate on translations from their mother tongue into a foreign language (Poland, 2019).

The professional reason for this is plain. When translating into the mother tongue, control over style comes naturally; translating into a foreign language, every sentence calls for extra verification, the work takes longer and a native-speaker reviser usually comes into the process.

Finding	Value	Label
Applying a higher rate when translating into a foreign language (non-CAT users)	80%	Poland, 2019
Standard page equivalent	1,800 characters ≈ 240 words	Book, p. 83
Seeing price as an obstacle (non-CAT users)	roughly 40%	Poland, 2019
Data on rate levels for Türkiye	No publicly available measurement	–

It needs stressing that this proportion was measured only in the group of non-users. The data does not say “those who use tools disregard the direction”; we simply cannot answer the same question with a figure for that group in this article.

For the buyer the finding turns into a practical question: on a translation into a foreign language, is a reviser whose mother tongue is the target language involved? If that is the justification for the difference in price, the process has to include it.

Repetition discounts: the tool’s direct effect on the rate

The most concrete effect of tool use on rates is the discount scale. A translation memory classifies the segments in a file by match rate, and a different rate is applied to each class.

The logic is this: paying the full rate for a sentence that has already been translated and repeats word for word is unnecessary. On partly similar sentences the labour falls but does not disappear; intermediate rates are used for that reason.

The problem in practice is that the scale is usually applied without being explained to the client. If it is not in writing which rate applies to which match band, the discount turns into a bargaining instrument and no comparison can be made.

One more warning: high-match segments do not always call for less labour. When the context changes, the translation of the same sentence can change with it. A discount applied blindly can leave the labour of review unpaid; the literature calls this “blind faith in translation memory” (Bowker 2005).

How the scale is built matters as well. Exact matches, context matches and fuzzy match bands correspond to different levels of labour. Quotations offering a single “repetition discount” rate hide that distinction and are therefore not suited to comparison.

Whose pocket does the cost of the tool come out of?

The invisible side of the debate about rates is the cost base. In the research, 91% of participants said they owned the licence for the tool they used themselves (Poland, 2019).

The cost has a time component too: moving to a new tool means learning time and the migration of the old memories. That time is not billed but it reduces capacity and enters the price indirectly. In short, the finding means that the investment stays with the party doing the work. The licence fee, the annual maintenance and the learning time are the translator’s costs, and sooner or later they are reflected in the unit rate. That roughly 40% of the non-users in the same group gave price as an obstacle bears out this burden.

What completes the equation is the expectation of a discount. The translator makes the investment and then hands part of the return back to the client as a discount. We took up the detail of this structure in our article on licensing.

For the buyer the conclusion is this: an extremely low price is most often achieved by cutting the budget set aside for tools, training and quality control. Asking where a difference in price comes from is as important as bargaining over it.

The reverse needs saying too: a high price is not on its own a guarantee of quality. Asking where the difference goes – is there revision, is terminology work being done, is a quality control report produced – is useful in both directions.

Did the research measure the level of rates?

We have to state one limit plainly here. The case study did not produce a comparative table of rate levels; we cannot pass on the answer to the question “what do you charge per word” as a figure in this article.

The verified finding we have on rates is the 80% figure on the difference in direction (Poland, 2019). The other inferences about rates are qualitative judgements resting on industry practice and the logic of discounts.

An indirect indicator can be added to that: the proportion making their living from translation alone is 75% among CAT users and 41% among non-users (Poland, 2019). That is a measurement of working arrangement rather than of income level, and it cannot be counted as evidence about rates.

Let us recall once more that the sample consists of 141 people, rests on voluntary participation and belongs to 2019. All of these constraints are set out in detail in our article on the limits of the research.

The consequence of limits of this kind that is most often missed is this: staying silent about a subject a survey did not measure is the same as saying there is no data on it. That is why we use qualitative wording rather than percentages in areas that have not been measured.

The variables that really set the rate

The tool is only one variable in the rate equation, and usually not the most decisive one. The elements that really build the price lie elsewhere.

Specialisation comes first among them: medical, legal and patent texts carry a risk of liability and draw on a narrower pool of specialists. Then come the rarity of the language pair, the delivery time, the format of the source file and the approval procedures.

The format of the source file should not be taken lightly. A scanned PDF calls for extra processing compared with an editable text file, and that processing is reflected in the rate. The same content can be priced differently in a different file format.

Tool use is a layer sitting on top of these variables. On jobs with a high repetition rate it brings the total down; on jobs with no repetition it barely touches the price at all.

Then there are the invisible items: project management, term approval, the processing of client feedback and the final review. When these steps are not shown separately in a quotation the price looks low, but the workload falls either on the buyer or on the quality.

The Türkiye context

There is no publicly available, comparable study measuring the level of translation rates in Türkiye. We therefore give no figure for Türkiye to set beside the Polish data.

On the question of units, however, there is a marked local difference. In the sworn translation market in Türkiye, pricing is mostly done by the page, and the definition of a page can vary from one agency to another. In corporate and technical translation, word-based pricing is widespread. Before two quotations can be compared, both have to be converted to the same unit.

The structure of Turkish comes directly into play too. Because the language is agglutinative, several words in an English sentence can come down to a single word in Turkish. Whether the count is made on the source language or on the target language can create a notable difference in the amount for the same job. This needs to be set down in writing in the quotation. We have explained our own unit structure on the pricing page.

Another local difference in practice is that tax and invoicing arrangements are not always presented the same way in quotations. When making a comparison, it is necessary to establish whether the amounts are inclusive or exclusive of tax.

What does this mean for you?

What you need to look at when comparing quotations is not the unit price but the total cost of the same job. A few steps are enough to make that visible.

Convert every quotation to the same unit; ask whether the count is made on the source or on the target.
If a repetition analysis is being run, ask in writing for the match bands and the rates applied to them.
Settle the question of direction: on a translation into a foreign language, is there a native-speaker reviser?
Ask whether extra processing arising from the file format is included in the price.

We give a quotation not as a single figure but together with the volume, the repetition analysis and the process items, so that what you are paying for and why becomes visible. If you would like this breakdown for your own file, you can use the quotation form.

Frequently asked questions

Does a translator who uses CAT work more cheaply?

The research gives no such general conclusion. Tool use does bring the expectation of a repetition discount with it, however; that can bring the total amount down, though not the unit rate.

Why is the rate different when translating into a foreign language?

Translating into the mother tongue is usually faster and safer. In the sample, 80% of the non-CAT users said they applied a higher rate when translating into a foreign language (Poland, 2019).

Why does the unit matter in a translation rate?

The same job can come to different amounts by the word or by the character. The book states that 1,800 characters with spaces correspond to roughly 240 words; do not compare quotations without converting them to the same unit.

Is it right to ask for a repetition discount?

If your documents really are repetitive, yes. Before asking for a discount, ask for the repetition analysis to be run and for the rate applied to each match band to be put in writing.

References

Pietrzak, P. & Kornacki, M. (2021). Using CAT Tools in Freelance Translation. Routledge Focus, pp. 83-84.
Bowker, L. (2005). Productivity vs quality? A pilot study on the impact of translation memory systems. Localisation Focus, 4(1), pp. 13-20.
DePalma, D. (2012). Translation Productivity Benchmarks. Common Sense Advisory.